Upcoming WebinarsMore Webinars
Alan Dix explores AI and Social Justice
Author of Introduction to Artificial Intelligence
AI Business is part of the Informa Tech Division of Informa PLC
This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 3099067.
The three key properties of 5G–ultra-reliable low latency communication, enhanced mobile broadband, and massive machine-to-machine communications–make for a compelling argument. These properties will be instrumental in IoT and, as such, many future smart home applications.
Future smart homes will rely on ultra-reliable low latency communications links that enable users, and especially devices, to be in constant real-time communication. While one of the three properties of 5G is ultra-reliable low latency, 5G won’t be the only wireless internet access in our homes in the immediate future–Wi-Fi will also remain in homes, as it is now, alongside a number of existing devices that only support 4G.
Many people already experience “connectivity conundrums” in their own homes, whether their Wi-Fi does not provide optimal service in the kitchen, or 4G offers a stronger connection in the living room. For most, the best internet service flits between cellular technologies and Wi-Fi, dependent on the users’ position, physical layout of the home, the number of physical barriers present (a problem particularly relevant to mmWave radio which supports 5G) and a user’s changing proximity to radio access points.
Even within today’s applications, manually switching between radio access points can create a negative user experience and somewhat impact their functionality. We might need to switch from Wi-Fi to 4G during conference calls to prevent the connection from dropping, but if we aren’t quick enough, we may be kicked off the call. And the applications of the future may have even more challenges in store.
Let’s consider the smart home application of a fully immersive and high resolution 3D gaming experience, which combines the real and virtual worlds in an augmented reality fashion. In this scenario, an optimal network service and the movements of the user are both imperative to successfully implement a fully-immersive 3D game where users can navigate a real world augmented with virtual objects in their own home. It simply won’t work to require users to manually handover the connection from one radio access to another, especially as connectivity quality often fluctuates based on the user’s movements within the game. Not only will it create a negative user experience, but without the best possible internet connectivity, latency could cost the user the game.
As we begin to imagine other applications in the future smart home–from holographic video calls to semantics-based interactive streaming TV services–it is vital that they are supported by optimal network service, regardless of the user’s position within their home or whether the service is offered via Wi-Fi or cellular technologies. Without optimal network service, real-time communication between humans and devices risks being broken.
This challenge with users manually switching between radio access points to capture optimum service is where AI steps up to the mark. The fusion of AI and wireless uses AI algorithms to select the best radio access point based on user movements and application demands and then automatically switches between radio access points, without user intervention, to maintain optimal connectivity.
Within wireless technologies, AI algorithms have been trained to detect, identify, and diagnose network connectivity issues by leveraging metrics at different levels in the communications layers, also known as Supervised Learning. It will then label the issue and report back to the user. While this is an important step forward, it is not new, and AI classification algorithms will not help in the gaming scenario above or other instances where the user is required to manually switch the devices’ radio access points.
While Supervised Learning relies on making assumptions about a large training data set as a prerequisite, Reinforcement Learning AI algorithms make very few assumptions on the initial settings, and instead learn on the go. The output wouldn’t be a classification or diagnosis, but rather an action or reaction type of operation. And with wireless technologies, this action could be switching a device’s connection from one radio access point to another.
Reinforcement Learning AI continually monitors a system, gathering metrics and building a state or ‘idea’ of the system and framework in which it is operating. Once it has gained an ‘understanding’ of the system, the algorithm will perform actions. Each time the algorithm acts, it also computes the effect, learning what actions, in what conditions, have what effect. Over time, the AI algorithm will train itself to perform the right actions, in the right circumstances, without any interaction from the user.
For the smart home of the future, this will be a transformative approach. Instead of users worrying, or even thinking, about what radio bearer will offer the best service, the reinforcement learning AI algorithms will automatically and seamlessly switch between those available–whether that’s 4G, 5G or Wi-Fi. This ensures users receive the best quality of service possible, regardless of their position, and enables them to experience smart home applications–such as immersive AR 3D gaming–at their full potential.
It’s clear that 5G and the cellular technologies that follow will be instrumental in enabling many smart home applications in the future. However, when it comes to providing the technological foundations, we must not underestimate the role of AI. Reinforcement Learning AI algorithms will ensure that applications we can only imagine today become applications that future generations cannot live without.
Author of Introduction to Artificial Intelligence